Lecture 12 Flashcards
3 uses of biochemical data in clinical medicine (and important aspects)
1) diagnosis : suspected, investigation, results compared to expected range -> OVERLAP between healthy and diseased range
2) management : assess disease severity (often correlated to extent of abnormality of result
3) monitor disease progression + response to treatment
what does screening for diseases imply ? example
attempting to detect disease before it manifests through the development of clinical disturbance
example : selective screening (neonatal)
what are the three steps in the testing process ?
1) pre-analytical : collect sample, deliver to lab, …
2) analytical : calibrate, quality control, measure
3) post-analytical : deliver results, act on them
6 analytical factors to take into consideration to minimize errors (and their definition)
1) analytical range : test has to cover range of concentrations expected
2) accuracy : ability to produce a result that reflects the true value
3) bias : difference between mean of a set of replicate measurements and the true value (systematic error, persistent)
4) precision : reflection of reproducibility (imprecision = standard deviation for a series of the same measure)
5) specificity : measure only the analyte that we want
6) interference : substance alters the signal of the analyte we want, but doesn’t generate a signal itself
what is the analytical goal ? (overall imprecision)
analytical imprecision (CVA) < 0.5*intra-individual variability (CVI)
-> overall : 1,12 * CVI
predictive value of a test : explain what TN, TP, FN, FP are (explain the graph)
TP = has disease, tests positive
TN = doesn’t have disease, tests negative
FP = doesn’t have disease, tests positive
FN = has disease, tests negative
graph is two bell curves that intersect (cut-off value)
5 important parameters relating to test performance (formulas)
1) prevalence : people with disease / total popu.
2) sensitivity : prob. of + result in a diseased individual (TP/TP+FN) -> intrinsic
3) specificity : prob of - result in healthy individual (TN/TN+FP) -> intrinsic
4) positive predictive value : prob. that + test is a diseased person (TP/TP+FP) -> not intrinsic
5) negative predictive value : prob. that - test is a healthy person (TN/TN+FN) -> not intrinsic
what happens for positive and negative predictive value if prevalence goes to 0 or to 1 ?
if prevalence -> 0 :
PPV -> 0
NPV -> 1
if prevalence -> 1 :
PPV -> 1
NPV -> 0
what is the efficiency of a test ? (formula)
fraction of individuals corectly assigned as either diseased or healthy
(TP + TN) / (TP + TN + FN + FP)
how do we determine the optimal cut-off ?
max distance to random classifier or closest to sensitivity = 1
what are the likelihood ratios ?
how much more likely a test result is to occur in an individual with a disease than without it (positive LR = sensitivity / 1 - specificity)
not to occur -> negative LR = 1 - sensitivity / specificity
what are the regulatory requirements for injectable ?
- safety
- sterility
- free from pyrogenic contamination
- free from visible particulate matter
- stability
- compatibility
- isotonicity
what kind of tests are in the routine microbial limit testing in industry ?
- bioburden assessment
- environmental testing
- raw materials testing
- process water testing
- sterility testing
- in-process testing
how is an environment test done ?
standing plate : 1m height, 1m from obstacle, 1 hour exposure time -> bacteria and fungi
formula to go from settle plate in CFU/cm2/hours to active air sampling in CFU/m3
N = 5a10^4*(bt)^-1
a = nb of colonies per plate
b = plate size cm2
t = exposure time
